OPTIMAL GRAPH SEARCH BASED SEGMENTATION OF AIRWAY TREE DOUBLE SURFACES ACROSS BIFURICATION Prepared by: Rosebud Roy Roll No:21,MTech CSE Guided by: Prof.Sreeraj R The Head of CSE Department

OBJECTIVE To do fully automated extraction on airways by optimal graph search based segmentation in MDCT images and to obtain accurate accessment of the inner and outer airway wall surface with less false positive rate.

INTRODUCTION The structure of lungs airway is of complex nature, particularly around the branch areas,so diagnosing disease based on thickness is difficult. We need to find accurate assessment of the inner and outer airway wall surfaces of a complete 3-D tree structure.[1][2] Numerous major lung diseases including chronic obstructive pulmonary diseases (COPD) and asthma are diagnosed by this thickness. X. Liu*, D. Z. Chen, M. H. Tawhai, X. Wu, E. A. Hoffman, and M. Sonka.”Optimal Graph Search Based Segmentation of Airway Tree Double Surfaces Across Bifurcations.” IEEE Trans ON MEDICAL IMAGING, VOL. 32, NO. 3, 2013

Illustrating complex airway tree structure. (a) Original MDCT image. (b) Rendered airway tree presegmentation.

OPTIMAL GRAPH SEARCH BASED SEGMENTATION[1][2] The graph search based algorithms solve the image segmentation problem by transforming it to finding a minimum cost closed set. Our graph search based image segmentation approach consists of the following four major steps.They are: 1. Presegmentation and meshed surface representation 2.Image resampling. 3.Graph construction 4.Graph Search X. Liu*, D. Z. Chen, M. H. Tawhai, X. Wu, E. A. Hoffman, and M. Sonka.”Optimal Graph Search Based Segmentation of Airway Tree Double Surfaces Across Bifurcations.” IEEE Trans ON MEDICAL IMAGING, VOL. 32, NO. 3, 2013

Presegmentation and meshed surface representation A presegmentation is needed to provide basic information on the object’s global topological structure. If the presegmentation does not yield a surface mesh, we also need to transform the volumetric result into a mesh representation. Image resampling Using the outcome of the presegmentation,the image is resampled based on each vertex of the initial surface mesh directly, resulting in a set of vectors (called columns) of voxels K. Li, Student Member, IEEE, X. Wu, Senior Member, IEEE,D. Z. Chen, Senior Member, IEEE, and M. Sonka, Fellow, IEEE.”optimal Surface Segmentation in Volumetric Images—A Graph-Theoretic Approach.” IEEE Trans ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, VOL. 28, NO. 1, 2006

Graph construction Each voxel in the columns is considered as a vertex in a graph. A cost is assigned to each vertex of the graph which reflects certain properties of the sought surfaces. Graph search A minimum - cut algorithm, is applied to the resulted graph to simultaneously search for multiple inter-related surfaces.

Fig: a)Column length too short b) column length high.Image resampling and graph) construction. (a1) Interferences caused by inappropriate column lengths introduce disordered structures in the constructed graph. (b1) 2-D example of image resampling based on medial axes to obtain columns with no interference..

METHODS

REGION GROWING METHODS [3] It is a type of segmentation used. Because of an imaging artifact (ultrasound image) or a thin airway wall, the region growing approach may allow the growing process to jump from the inside of the airway to the pulmonary parenchyma,which frequently carries similar gray-level properties on CT images. Once the growth starts outside of an airway lumen (growing “leaks”), there is nothing to stop it and large parts of the lungs are erroneously marked as the airway tree. Mori K, Suenaga Y, Toriwaki J. “Automated anatomic labeling of the bronchial branch and its application to the virtual bronchoscopy.” IEEE Trans Med Imaging,VOL. 36,NO. 2, 2000;19:103–114.

Contd.. In the case of low-dose scans, the segmentation either stops early or leaks. Region-growing algorithm was run 12 times on every dataset to find an optimal combination of parameters. Limitation: It can't segment lowdose scans and high execution time.

FUZZY CONNECTIVITY METHOD[4] It is a another airway segmentation method based on fuzzy connectivity. During the execution of this algorithm, growth of the foreground region and growth of the background region compete against each other. This method has the great advantage that it can overcome lack of image contrast between the airways and the airway walls and the effects of noise. J. Tschirren, E. A. Hoffman, G. McLennan, and M. Sonka, “Segmentation and Quantitative Analysis of Intrathoracic Airway Trees from Computed Tomography Images” Departments of Electrical and Computer Engineering, Radiology, Internal Medicine, and Biomedical Engineering, University of Iowa,Iowa City, Iowa

OPTIMAL GRAPH SEARCH SEGMENTATION In optimal graph search, a segmentation can be done with accuracy in CT scans.Below advantages are there compared to previous ones.[1] low false positive rate Accuracy will be high less execution time no leakage and noise fully automatic can do in CT Scans

CONCLUSION This method now allows for the separate measurements of the airway wall thickness in the bifurcation and carina regions of the airway trees using a globally optimal approach. The quantification of wall properties in bifurcations offers an effective basis for novel disease-specific studies of the intrathoracic airway tree morphology and function. Our method is general and can be applied to segmenting other complex objects with multiple inter-related surfaces in 3-D and higher-D images.[10]

FUTURE WORK To access the accurate thickness of outer and inner walls of airways without leakage and fully automatic in CT and MDCT scans.Also trying for detecting Malignous nodule. This segmentation has to work with low false positive rate. The extraction method must work within less execution time.

REFERENCES [1] X. Liu*, D. Z. Chen, M. H. Tawhai, X. Wu, E. A. Hoffman, and M. Sonka.”Optimal Graph Search Based Segmentation of Airway Tree Double Surfaces Across Bifurcations.” IEEE Trans ON MEDICAL IMAGING, VOL. 32, NO. 3, 2013 [2] K. Li, Student Member, IEEE, X. Wu, Senior Member, IEEE,D. Z. Chen, Senior Member, IEEE, and M. Sonka, Fellow, IEEE.”optimal Surface Segmentation in Volumetric Images—A Graph-Theoretic Approach.” IEEE Trans ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, VOL. 28, NO. 1, 2006 [3] Mori K, Suenaga Y, Toriwaki J. “Automated anatomic labeling of the bronchial branch and its application to the virtual bronchoscopy.” IEEE Trans Med Imaging,VOL. 36,NO. 2, 2000;19:103–114.

[4]J. Tschirren, E. A. Hoffman, G. McLennan, and M. Sonka, “Segmentation and Quantitative Analysis of Intrathoracic Airway Trees from Computed Tomography Images” Departments of Electrical and Computer Engineering, Radiology, Internal Medicine, and Biomedical Engineering, University of Iowa,Iowa City, Iowa [5]Z. Xu, U. Brent, F. Daniel, J. Mollura, “A hybrid multi-scale approach to automatic airway tree segmentation FROM CT SCANS”. Center for Infectious Disease Imaging (CIDI) Radiology and Imaging Science Department National Institutes of Health (NIH), Bethesda, MD [6]Law TY, Heng PA. “Automated extraction of bronchus from 3D CT images of lung based on genetic algorithm and 3D region growing.”Proc SPIE Int Soc Opt Eng 2000;3979:906– 916.

THANK YOU..